OpenAI says research org runs 3.1 agent-workdays per human workday
Cklly2004 · reddit · 2026-09-08
OpenAI published new internal data on how its researchers use coding agents:
- The research organization now uses 3.1 agent-workdays of effort per human workday, while researchers write more code and run more experiments than before
- Delegated work has grown more complex, reaching the goal of an "automated research intern" handling well-defined tasks that would take a skilled researcher several days
- Humans still set priorities and judge results; over half of successful 4–8 hour tasks required at least one human intervention
The post ends by asking whether humans are becoming research managers while agents provide most execution.
Related event: OpenAI Data: Most Long-Hour Agent "Successes" Required Human Help(2 posts)→
More from AGI Musings
- Nat Lambert: open models deserve attention more than frontier lab drama — deliprao · 2026-09-08
- Luiza Jarovsky: it's time to hire human photographers and designers again — LuizaJarovsky · 2026-09-08
- Terence Tao: premature AI solutions to math problems could harm mathematics itself — ValerioCapraro · 2026-09-08
- TheZvi: Math proofs are starting as giant AI output piles—will first versions ever be comprehensible? — TheZvi · 2026-09-08
- Skeptic: AI slowdown pledges are really about blocking rivals, not safety — yunta_tsai · 2026-09-08
- Office buildings will empty out between end of 2027 and 2030, predicts OpenAI insider — davidpattersonx · 2026-09-08